Understanding Society User Support

Good evening,

I have some brief queries about above variables.

By way of background (and mostly just to provide some detail for other readers): I note that from wave 2-onwards, Understanding Society has a family of health variables in the self-completion questionnaire giving lots of detail on the way peoples' health is limited (all these variables have the "scsf" prefix). I also note that the same questions were directly asked of respondents, ie not in the self-completion questionnaire, in wave 1 (this data being prefixed "sf"). I can see that the general health question (called "sf1") continues to be asked directly of respondents in addition to also being asked in the self-completion survey (where it's called "scsf1"), but apart from that, from wave 2-onwards all the more detailed questions are only asked in self-completion.

I note that there are two very useful derived variables which take all these answers and convert them into a 0-100 scale, using a well-established methodology - one for the mental health-related questions (sf12mcs_dv); and another for the physical health ones (sf12pcs_dv). I am interested in using this data from Wave 4, 6, 8, 10 and 12 of Understanding Society for a study I'm working on, although I may also use some of the more detailed "scsf" questions for some analysis.

I have a couple of questions about this data:

1. I can see that there are far more inapplicable (-8) and proxy (-7) missing values in earlier waves of the data. Eg 16.5% of respondents are recorded as -8 or -7 in Wave 4, but only 3.5% are in Wave 12. Could you shed any light on why this is, particularly for the inapplicables? I have checked the question routing in the user survey questionnaire and they appear to be exactly the same in Wave 4 and Wave 12, hence my confusion.

2. Can you confirm that from wave 2-onwards, the derived variables (sf12mcs_dv / sf12pcs_dv) are derived exclusively from the self-completion data, and there isn't a direct set of questions I'm not aware of? The UKHLS variable search webpage still refers to "sf" rather than "scsf." I ask chiefly because presumably this affects the weight to be used, as I understand I would need to switch to a "_indsc" weight if I add this derived data to my analysis.

Best wishes,

Tom